[Paper Review] Tractable Resource Management in Millimeter-Wave Overlaid Ultra-Dense Cellular Networks

Jihong Park, Seong‐Lyun Kim+1 more|arXiv (Cornell University)|Jul 16, 2015

Advanced MIMO Systems Optimization14 references3 citations

TL;DR

This paper proposes a tractable resource management framework for millimeter-wave (mmWave) overlaid ultra-dense networks (UDNs), leveraging stochastic geometry and a 3D blockage model to derive closed-form spectral efficiency (SE) expressions. It finds that micro-wave (μW) spectrum should be predominantly allocated to uplink (UL) to counteract mmWave's high peak-to-average power ratio (PAPR), which severely limits UL rates, and that optimal resource allocation depends logarithmically on base station (BS) density ratios.

ABSTRACT

What does millimeter-wave (mmW) seek assistance for from micro-wave (μW) in a mmW overlaid 5G cellular network? This paper raises the question of whether to complement downlink (DL) or uplink (UL) transmissions, and concludes that μW should aid UL more. Such dedication to UL results from the low mmW UL rate due to high peak-to-average power ratio (PAPR) at mobile users. The DL/UL allocations are tractably provided based on a novel closed-form mm-μW spectral efficiency (SE) derivation via stochastic geometry. The findings explicitly indicate: (i) both DL/UL mmW (or μW) SEs coincidentally converge on the same value in an ultra-dense cellular network (UDN) and (ii) such a mmW (or μW) UDN SE is a logarithmic function of BS-to-user density ratio. The corresponding mm-μW resource management is evaluated by utilizing a three dimensional (3D) blockage model with real geography in Seoul, Korea.

Motivation & Objective

• Address the inherent uplink (UL) rate bottleneck in mmWave-overlaid 5G ultra-dense networks (UDNs), caused by high peak-to-average power ratio (PAPR) at mobile devices.
• Investigate how incumbent micro-wave (μW) spectrum can be optimally allocated between downlink (DL) and UL to maximize overall DL rate while ensuring a minimum UL rate.
• Develop a tractable, closed-form resource management solution using stochastic geometry and realistic 3D blockage modeling to guide network design and deployment.
• Provide design guidelines for 5G mmWave-μWave heterogeneous networks by quantifying the impact of BS density and spectrum allocation on spectral efficiency.

Proposed method

• Model the network as a two-tier heterogeneous system with mmWave and μWave base stations (BSs) distributed as independent 2D homogeneous Poisson point processes (PPPs).
• Introduce a 3D blockage model based on real geographic data from Seoul, Korea, to capture line-of-sight (LOS) propagation constraints for mmWave signals.
• Derive closed-form expressions for downlink and uplink spectral efficiencies (SEs) of both mmWave and μWave links using stochastic geometry and tractable approximations.
• Apply upper and lower bounds to derive tractable expressions for UL SE, enabling analytical optimization under constraints.
• Formulate a resource allocation problem to maximize DL rate under a minimum UL rate constraint and a PAPR outage probability limit.
• Solve the optimization problem in closed form, yielding an optimal μWave UL resource allocation policy dependent on BS densities and spectral bandwidths.

Experimental results

Research questions

• RQ1 How should micro-wave (μW) spectrum be allocated between downlink (DL) and uplink (UL) in mmWave-overlaid ultra-dense networks to maximize overall DL rate?
• RQ2 What is the impact of mmWave’s high PAPR on uplink spectral efficiency, and how can this bottleneck be mitigated using μWave resources?
• RQ3 How do base station densities and bandwidth allocation affect the spectral efficiency of mmWave and μWave links in a 3D blockage environment?
• RQ4 Can closed-form expressions for spectral efficiency be derived in ultra-dense networks with realistic 3D blockages to enable tractable optimization?
• RQ5 What is the optimal trade-off between DL and UL resource allocation when the minimum UL rate is constrained?

Key findings

• For a 500 MHz mmWave bandwidth and a 20% minimum UL/DL rate ratio, over 60% of the μWave spectrum should be allocated to UL to resolve the UL bottleneck.
• With 1 GHz mmWave bandwidth, the entire μWave spectrum should be dedicated to UL to meet the minimum UL rate requirement.
• The spectral efficiency of both mmWave and μWave links in ultra-dense networks increases logarithmically with the ratio of BS-to-user density.
• The derived closed-form expressions for downlink and uplink spectral efficiencies are validated using a 3D blockage model based on real building data from Seoul, Korea.
• The optimal μWave UL resource allocation increases with mmWave bandwidth and required UL rate, contradicting current trends favoring DL resource dominance.
• The maximum achievable downlink rate scales logarithmically with the product of mmWave and μWave BS densities, confirming the benefit of network densification.